Alpha
Contact
Assess · Berget AI

Can you own it?

Ownership levelSubstantialnone·limited·partial·substantial·fullAnalytical input C ยท 67.2/100

This page is a projection of the one entry record, the Use & modify and Transparency factors that Assess covers. The full verdict is set by all four factors together, floor-weighted so the weakest caps the whole.

Which domain expands which factor
  • AssessUse & modify + Transparency
  • ImplementData control + Reliability
  • UseReliability
  • SupportTransparency

Data governance - retention, ZDR & training

Zero data retention is the default for serverless inference, stated in two binding documents. The Data Processing Agreement (updated 2026-10-22, and first in the Terms' order of precedence) says in Appendix 2 that Berget applies "a true Zero Data Retention policy", stores no prompts, files or outputs after completed calls, and keeps none of it on persisted disks even during inference. The Terms of Service (Sec 7) say Berget will never store the actual prompt content or the output content. The DPA also has an express no-training sentence: "Berget AI does not train any models on the input or output from serverless inference endpoints, nor does it share the data with any model providers for model training." The catalogue is open-weight only, so no closed third-party model's data terms attach.

Inference retention is separate from account termination. On termination, Customer Data is kept 30 days for export and recovery, then irreversibly deleted. Berget does keep metadata (service logs, API request metadata, token counts, API key and subscription ID for billing). Two caveats apply. The Terms reserve a right to aggregated, anonymised data that identifies neither the customer nor any natural person (Sec 19.1). The retention guarantees are not independently attested: no SOC 2 or ISO 27001 audit stands behind them, which is why data governance is held at 4, not 5.

Data & IP ownership

Your data stays yours, per the Terms. As between Berget and the customer, the customer retains all title and intellectual property rights in Customer Data (Sec 17.1). Berget's licence is scoped to hosting, using, processing and transmitting it to provide the Service. Sec 18 is a mutual confidentiality clause whose definition includes "in particular Customer Data", with a duty on termination to cease use and destroy (Sec 18.5). The DPA names four sub-processors (AI Iron AB, Acebit AB, Obenet AB and Stripe Payments Europe, Limited) and gives 30 days notice of changes with an objection and termination right. The caveats are that the assurance rests on binding terms with no independent audit, and the Sec 19.1 aggregated-data reservation. The customer feedback licence (Sec 19.2) covers suggestions only.

Data residency & jurisdiction

The binding boundary is the EEA, and the contract does not promise Sweden. DPA Sec 10.1 says all Berget data centres are in the EEA, and Sec 10.2 says Berget will not transfer personal data outside the EEA without the customer's written instruction. The company is Swedish-owned with no US parent, which keeps it outside US CLOUD Act reach. Three points temper the marketing. First, Sec 10.1 lets the customer choose a data centre only for compute and storage, "not including serverless inference", so a serverless customer cannot pin a data centre. Second, the statements that data "stays within Sweden's borders", "never crosses borders" and involves "no US entity" appear on the landing page, in the docs and in the Berget Code post, and no clause promises them. Third, the Privacy page is a website privacy policy and says nothing about where inference runs. The payment sub-processor, Stripe Payments Europe, Limited, is in Ireland. For an EU adopter this is a strong jurisdictional position, scored 4 because the main workload has no region pinning.

Compliance & attestations

Berget publishes a GDPR Data Processing Agreement at berget.ai/en/dpa. It is an Art. 28(3) controller-processor agreement with named sub-processors, a 24-hour breach notice where feasible, an audit right (one per 12 months, customer pays), a 30-day return or deletion on termination, and a Chapter V transfer clause (EU SCCs where applicable). Formal security attestations - SOC 2, ISO 27001, ISO 27701, HIPAA/BAA - do not exist on any Berget page. ToS Sec 6.7 says Berget "is actively working toward ISO 27001 certification". For ISO-dependent or healthcare procurement, treat compliance as "GDPR DPA in place, no certificate yet".

Security controls

ToS Sec 6.7 states industry-standard technical and organisational measures - encryption in transit, network isolation, access controls and system monitoring - and DPA Sec 8 adds pseudonymisation and encryption, restore capability and regular testing of the measures. The page titled Security is a responsible-disclosure policy (GitHub private reporting, a PGP key, a Signal number, a typical 48-hour response and a safe-harbour statement) and is not a trust centre. No independently reviewed penetration test or SOC 2 / ISO 27001 attestation exists, so the posture is stated but untested.

Pricing & cost model

Per-model per-token rates are published through the API listing, in euros per million tokens: for example Mistral-Small-3.2 0.30 in and 0.30 out, Gemma 4 31B 0.25 and 0.50, Kimi K3.1 Pro 3.00 and 10.00. There is no pure pay-as-you-go: the pricing FAQ says API access needs an active plan, from the free Trial (EUR 5 credits) through Starter EUR 25, Team EUR 50 and Enterprise EUR 500 per month, each a prepaid top-up. Credits are non-refundable and expire after 12 months. Berget Code is EUR 150 to 500 per seat per month, and Dedicated Inference is quoted from about EUR 2,500 per GPU per month. Zero retention is the default and carries no premium.

Reliability posture

The SLA (2026-10-23) gives serverless inference, including Berget Code, a best-effort level with "no compensation payable for interruptions". A 99.5% monthly uptime commitment with fee credits of 10% or 25% covers only Dedicated Inference with an HA setup, which is sales-contracted. The SLA names status.berget.ai for incident information, but that page is a JavaScript shell with no readable content, so no incident history or uptime record is available. The tech.eu article supports production use since 2025 but names no customers. Serverless is unsuitable for availability-critical workloads without a dedicated contract.

How this scores

The ownership factors this domain covers, drawn from the one entry record.

1

Use and modify freelyCan you use it freely and leave without lock-in?

Strong

OpenAI-compatible API (api.berget.ai/v1) over open-weight models (Gemma 4, Mistral-Small, GLM-5.3-Flash, Kimi K3.1 Pro) - migrate by switching base URL and key, and the same checkpoints run elsewhere or self-hosted; the lifecycle policy retires Active models with 2 to 3 weeks notice.

How this scores (AOI sub-dimensions)
Transparency & lock-in5/5how portable it is and how easily you can leaveOpenAI-compatible API over open-weight models with published licences and a published lifecycle policy (docs.berget.ai: 2 to 3 weeks notice for Active models, named successor, retired-model list), independently listed by the Mastra model directory as an OpenAI-compatible /chat/completions provider at api.berget.ai/v1.
Cost4/5how the pricing model compares and how predictable it isPer-model per-token rates are published through the API listing, plans are public (EUR 0 to 500 per month) and Dedicated Inference has an indicative starting price.
2

TransparencyAre the binding terms published, legible and independently checkable?

Moderate

The binding ToS, DPA, SLA and the Privacy page are published and legible, and the DPA (first in precedence) states ZDR, no training, EEA data centres and four named sub-processors. Held below strong because no SOC 2 or ISO 27001 attestation independently corroborates the guarantees, and the marketing overstates the binding text (Sweden-only, 'never crosses borders').

How this scores
Not a scored AOI dimension. For a hosted provider, transparency is whether the binding terms are published, legible and were actually read - the read/unverified evidence below, not a certification. A strong rating here must trace to a retrieved binding document.
What this means for adoptionYou substantially own workloads here. The binding DPA and Terms state zero retention of prompts and outputs, an express no-training sentence, customer-retained ownership under a service-scoped licence, a mutual confidentiality clause and a 30-day export window. An EEA-only data-centre boundary and Swedish ownership keep you outside US CLOUD Act reach, and a portable open-weight exit keeps switching cheap. Verification depth and operating terms hold it below full ownership. No independent audit or SOC 2 or ISO 27001 attests the guarantees, and serverless inference is best effort with no SLA credits. The contract promises EEA data centres with no data-centre choice for serverless, and Sweden-only is marketing. Adopt as an EU-based default for non-critical availability, and contractualise Dedicated Inference if you need an uptime commitment.

Sources

The same evidence records as the entry sheet. Read means the text was verified; unverified means it is known to exist but not yet read.

Terms of serviceread2026-11-19
Berget AI Terms of Service (page stamp 2026-10-16).
Privacy Policyread2026-11-19
Berget AI Data Privacy Policy (last updated 2025-03-10): a website privacy policy for visitors, listing website data, cookies and website retention periods.
Securityread2026-11-19
Berget AI Security page: a "Responsible Disclosure" policy with GitHub private vulnerability reporting, security@berget.ai with a PGP key and a Signal number, a typical 48-hour response, a scope list and a safe-harbour statement.
Data Processing Addendumread2026-11-19
Berget AI Data Processing Agreement (last updated 2026-10-22), an Art.
Slaread2026-11-19
Berget AI Service Level Agreement (last updated 2026-10-23), part of the Agreement under ToS Sec 3.1.
Vendor announcementread2026-11-19
Berget AI landing page (marketing): "Open models running under Swedish jurisdiction with zero data retention", "never stores the content of your prompts or generated output, and never trains on your data", "EU-only infrastructure, no US entity involvement".
Documentationread2026-11-19
Berget AI developer documentation (including llms.txt and Markdown pages): OpenAI-compatible chat completions API at https://api.berget.ai/v1 and a model lifecycle policy (Preview models removable without notice; Active models retired with 2 to 3 weeks notice and a named successor).
Documentationread2026-11-19
Berget AI live model listing (GET https://api.berget.ai/v1/models): 15 entries, priced in EUR per million tokens.
Third-party analysisread2026-11-19
Independent Mastra model directory lists Berget.AI as a provider reached through the OpenAI-compatible /chat/completions endpoint at https://api.berget.ai/v1, with 6 models routable.
Third-party analysisread2026-11-19
Independent EU tech press (Tech.eu, 2026-02-02): Berget AI raised EUR 2.1 million (SEK 24 million) in a round led by Luminar Ventures with Wellstreet and Norrsken Evolve.
Vendor announcementread2026-11-19
Berget AI pricing page (the per-model table is JavaScript-rendered, so rates come from ev-catalogue).
Vendor announcementread2026-11-19
Berget Code launch post (2026-05-13): agentic coding assistants.
Vendor announcementread2026-11-19
Berget AI Dedicated Inference page: reserved GPUs in Swedish data centres for a chosen or customer-supplied model, sales-led ("Talk to us").
Slaunverified2026-11-19
Berget AI status page.